Quantitative Assessment of Future Environmental Changes in Hydrological Risk Components: Integration of Remote Sensing, Machine Learning, and Hydraulic Modeling

Abstract

Floods are one of the most devastating natural hazards that have intensified due to land use land cover (LULC) changes in recent years. Flood risk assessment is a crucial task for disaster management in flood-prone areas. In this study, we proposed a flood risk assessment framework that combines flood vulnerability, hazard, and damages under long-term LULC changes in the Tajan watershed, northern Iran. The research analyzed historical land use change trends and predicted changes up to 2040 by employing a Geographic Information System (GIS), remote sensing, and land change modeling. The flood vulnerability map was generated using the Random Forest model, incorporating historical data from 332 flooded locations and 12 geophysical and anthropogenic flood factors under LULC change scenarios. The potential flood damage costs in residential and agricultural areas, considering long-term LULC changes, were calculated using the HEC-RAS hydraulic model and a global damage function. The results revealed that unplanned urban growth, agricultural expansion, and deforestation near the river downstream amplify flood risk in 2040. High and very high flood vulnerability areas would increase by 43% in 2040 due to human activities and LULC changes. Estimated annual flood damage for agriculture and built-up areas was projected to surge from USD 162 million to USD 376 million and USD 91 million to USD 220 million, respectively, considering 2021 and 2040 land use change scenarios in the flood-prone region. This research highlights the importance of land use planning in mitigating flood-associated risks, both in the studied area and other flood-prone regions.